This paper discusses the prediction of compression index (CC) on Brazilian coast soft soils by empirical correlations and artificial neural networks. The data from 295 standard consolidation (or oedometer) tests and the corresponding index properties performed on cohesive soils from different deposits of Brazilian coast are investigated herein. The predictions of alternative new empirical linear correlations relating CC with soil index properties for the investigated soft soils are compared with those of previous published CC empirical correlations. This paper also illustrates the use of artificial neural network (ANN) for CC predictions. The empirical correlations and ANNs performances evaluated through statistical techniques that include: (i) the root mean square error (RMSE), (ii) the ratio of the estimated measured compression index (K), (iii) the ranking index (RI) and (iv) the ranking distance (RD). The results indicate ANN as a potential alternative to empirical correlations for CC prediction of soft soils from Brazilian coast.